CN102156577A - Method and system for realizing continuous handwriting recognition input - Google Patents

Method and system for realizing continuous handwriting recognition input Download PDF

Info

Publication number
CN102156577A
CN102156577A CN 201110076004 CN201110076004A CN102156577A CN 102156577 A CN102156577 A CN 102156577A CN 201110076004 CN201110076004 CN 201110076004 CN 201110076004 A CN201110076004 A CN 201110076004A CN 102156577 A CN102156577 A CN 102156577A
Authority
CN
China
Prior art keywords
stroke
buffer area
historical
new input
stored historical
Prior art date
Application number
CN 201110076004
Other languages
Chinese (zh)
Other versions
CN102156577B (en
Inventor
何婷婷
葛勇
翟吉博
刘庆峰
胡郁
胡国平
王智国
Original Assignee
安徽科大讯飞信息科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 安徽科大讯飞信息科技股份有限公司 filed Critical 安徽科大讯飞信息科技股份有限公司
Priority to CN 201110076004 priority Critical patent/CN102156577B/en
Publication of CN102156577A publication Critical patent/CN102156577A/en
Application granted granted Critical
Publication of CN102156577B publication Critical patent/CN102156577B/en

Links

Abstract

The invention relates to the technical field of handwriting input and discloses a method and a system for realizing continuous handwriting recognition input. The method comprises the following steps of: tracking a stroke input chirography of a user and recording each stroke as a two-dimensional coordinate point column independently; determining a subordination relation between a new input stroke and the historic strokes stored in a first cache region; if the new input stroke and the historic strokes stored in the first cache region are different characters, emptying a second cache region and storing the historic strokes stored in the first cache region into the second cache region, and emptying the first cache region and storing the new input stroke into the first cache region; otherwise, storing the new input stroke into the first cache region; recognizing the historic strokes in the second cache region according to the two-dimensional coordinate point column of the recorded historic strokes so as to obtain a complete character; and displaying the recognized complete character. By the method, the input efficiency and the recognition rate can be improved effectively.

Description

Realize the method and system of continuous handwriting recognition input

Technical field

The present invention relates to the writing technology field, particularly a kind of method and system that realize continuous handwriting recognition input.

Background technology

In recent years, along with the fast sale of hand-hold electronic equipments, its function just progressively turns to from the simplest electronic address information book and has the multimedia terminal equipment that touches the microcomputer that shows and support infinite network.Along with the text on these mini-plants input demand from brief information, turn to longer notes or email message as electronic address book etc., efficient and continuous naturally input demand just is being subjected to increasing attention.

Giant-screen equipment is because to write the zone more wide in range, and the user is continued the writing style of daily continuous input.Yet on the small-sized touch panel device of portability, because the restriction of touch-screen size causes the person's handwriting of single input limited, the handwriting input ability is restricted continuously, and input efficiency is lower.

At this situation, the hand-written major part of supporting on this class small screen device is based on the monocase pattern at present, that is to say.User's in definite writing distinguished (as writing frame that presets or full frame) word for word imports, and the waiting system feedback of starting writing after individual character finishes empties then and writes the input that screen continues next character.This monocase input pattern has been avoided the character cutting difficult problem of continuous handwriting recognition, often has high recognition.Yet such input mode does not meet the daily writing style of people, and start writing wait in turn and etc. the input efficiency that influenced to be identified.

Obviously the hand-written pattern based on small screen device has much room for improvement, press in functionalization, hommization, aspect intelligent and make a breakthrough, as a kind of really available, easy-to-use, practical technology, be applied even more extensively in daily life and commercial career field.For this reason, proposed a solution in the prior art, this scheme allows the user at the predefined a plurality of characters of the continuous input of the mode that covers with stack in the zone of writing, and is implemented in limited continuous input demand of writing in the zone.But still there is following shortcoming in this scheme aspect user writing experience and the recognition performance:

At first, this scheme has adopted the continuous recognition technology of no cutting, just starts rear end identification after the user finishes all characters inputs, because needs etc. are to be identified, so input efficiency is subjected to certain influence.And, be directly on original space stack person's handwriting, to extract feature because prior art stack covers input mode, for fear of of the influence of space overlap information to character recognition, extraction all be stroke direction feature with location independent.And according to the traditional handwriting recognition technology, position feature to character particularly some easily to obscure the separating capacity of word stronger, the disappearance of position feature can influence the discrimination to input character.

Summary of the invention

The embodiment of the invention provides a kind of method and system that realize continuous handwriting recognition input, when realizing input continuously to solve the coverage mode that passes through in the prior art to superpose, and the problem that input efficiency and discrimination are low.

For this reason, the embodiment of the invention provides following technical scheme:

A kind of method that realizes continuous handwriting recognition input comprises:

Follow the tracks of user's stroke input person's handwriting, and each stroke is recorded as a two-dimensional coordinate point range separately;

Determine the membership of the new input stroke and the first buffer area stored historical stroke;

If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then empty second buffer area and store the first buffer area stored historical stroke into second buffer area, then, also will newly import stroke stores first buffer area into to empty first buffer area;

Otherwise, will newly import stroke and store first buffer area into;

Historical stroke in described second buffer area is discerned according to the two-dimensional coordinate point range of this history stroke of record, obtained complete character;

Show the complete character that identification obtains.

A kind of system that realizes continuous handwriting recognition input comprises:

Data acquisition module is used to follow the tracks of user's stroke input person's handwriting, and each stroke is recorded as a two-dimensional coordinate point range separately;

The breakpoint analysis module is used for determining the new membership of importing the stroke and the first buffer area stored historical stroke;

The stroke sorting module, be used for when described breakpoint analysis module determines that the new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters, empty second buffer area and store the first buffer area stored historical stroke into second buffer area, then, also will newly import stroke stores first buffer area into to empty first buffer area; And when described breakpoint analysis module determines that the new input stroke and the first buffer area stored historical stroke are under the jurisdiction of same character, will newly import stroke and store first buffer area into;

Character recognition module is used for the historical stroke of described second buffer area is discerned according to the two-dimensional coordinate point range of this history stroke of record, obtains complete character;

The character display module is used to the complete character that shows that described character recognition module identification obtains.

The method and system of the continuous handwriting recognition input of the realization that the embodiment of the invention provides, when realizing input continuously by the stack coverage mode in the prior art, adopt the low problem of continuous recognition technology input efficiency of no cutting, by following the tracks of user's stroke input person's handwriting, each stroke is recorded as a two-dimensional coordinate point range separately, in following the tracks of user's input process, by the breakpoint analysis mode the original character cutting of writing space stack is word string independently, and this word string is discerned synchronously.Because the processing of breakpoint analysis and character recognition is carried out synchronously with user's input, promptly in user's input process, determine that new input stroke is gentle and deposit that this carried out character recognition to historical stroke when historical stroke was under the jurisdiction of kinds of characters in the district, thereby avoided the stand-by period of waiting in turn under the existing stack covering input mode, improved input efficiency.And, owing to, can discern, therefore, can improve discrimination effectively to the historical stroke set that is judged as independent character by breakpoint analysis.

Further, handle by the desalination to historical stroke, can make the user clearly distinguish the stroke of historical person's handwriting and current input, the content that the user is imported clearly represents to the user.

Description of drawings

In order to be illustrated more clearly in technical scheme of the invention process, to do to introduce simply to the accompanying drawing of required use among the embodiment below, apparently, accompanying drawing in describing below only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain other accompanying drawing according to these accompanying drawings.

Fig. 1 is the process flow diagram that the embodiment of the invention realizes the method for continuous handwriting recognition input;

Fig. 2 is a process flow diagram of following the tracks of user's stroke input person's handwriting in the embodiment of the invention;

Fig. 3 is a kind of structural representation that the embodiment of the invention realizes the system of continuous handwriting recognition input;

Fig. 4 is the another kind of structural representation that the embodiment of the invention realizes the system of continuous handwriting recognition input;

Fig. 5 is a user writing stroke example in the embodiment of the invention;

Fig. 6 is writing the display effect example of district to stroke writing in the embodiment of the invention.

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the invention, the technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that is obtained under the creative work prerequisite.

The embodiment of the invention realizes the method for continuous handwriting recognition input, when realizing input continuously by the stack coverage mode in the prior art, adopt the low problem of continuous recognition technology input efficiency of no cutting, by following the tracks of user's stroke input person's handwriting, each stroke is recorded as a two-dimensional coordinate point range separately, in following the tracks of user's input process, will originally write character cutting that the space superposes by the breakpoint analysis mode and be word string independently, and this word string will be discerned synchronously.

As shown in Figure 1, be the process flow diagram that the embodiment of the invention realizes the method for continuous handwriting recognition input, may further comprise the steps:

Step 101 is followed the tracks of user's stroke input person's handwriting, and each stroke is recorded as a two-dimensional coordinate point range separately.

Wherein, each stroke is meant from starting to write a little to the continuous style of writing track of lifting between the pen point.

Step 102 is determined the new membership of importing the stroke and the first buffer area stored historical stroke.

Before and after two strokes writing may belong to same character and also may belong to two different characters, in embodiments of the present invention, can determine membership between different strokes according to the correlativity of stroke lettering feature and front and back stroke.

Particularly, can determine the membership of the new input stroke and the first buffer area stored historical stroke according to the position relation of stored historical stroke in the new input stroke and first buffer area.

At first,, calculate the start position of new input stroke according to the two-dimensional coordinate point range of stroke of record, and the final position of up-to-date stroke in the set of the first buffer area stored historical stroke, following several situation is arranged:

(1) distance in described start position and described final position is less than first distance threshold, such as, first distance threshold can be made as 0.2 times word frame length.Here word frame size can be adaptive to the user writing custom, it is set to be equivalent to write the numerical value of district's size when system initialization is not imported, many strokes according to user's input can recomputate word frame size subsequently, determine that then the described new input stroke and the first buffer area stored historical stroke are to be under the jurisdiction of same character;

(2) distance in described start position and described final position is greater than the second distance threshold value, such as, first distance threshold can be made as 0.35 times word frame length, determine that then the described new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters;

(3) distance in described start position and described final position is more than or equal to first distance threshold and smaller or equal to the second distance threshold value, the membership of then determining the described new input stroke and the first buffer area stored historical stroke is a fringe, that is to say, only can't judge accurately according to the position relation of stored historical stroke in the new input stroke and first buffer area whether new input stroke and historical stroke belong to same character.

Belong to the situation of fringe for this membership, can do auxiliary judgment further, particularly, can handle in the following manner:

A. the historical stroke set back that joins in first buffer area according to new input stroke is judged the influence of the equilibrium degree of character stroke on space distribution: make the equilibrium degree added value of accumulative total stroke surpass certain predetermined value if newly import stroke, can determine that then the historical stroke of newly importing in the stroke and first buffer area is under the jurisdiction of same character; Otherwise determine that the historical stroke in the new input stroke and first buffer area is under the jurisdiction of kinds of characters.

Described stroke distributing equilibrium degree is used to weigh the distribution of all sections (referring to the air line distance between the sampled point) in each interval of writing frame of stroke.At this, the distributing equilibrium degree of all strokes after need calculating the distributing equilibrium degree of historical stroke respectively and adding new stroke.Such as, at first calculate the word frame that stroke set to be investigated closes (i.e. all strokes in first buffer area), and this word frame is divided into four word frames, be respectively upper left quarter, lower left quarter, upper right quarter and right lower quadrant.Calculate subsequently wait to investigate accumulation the segment length of stroke on these four spaces and.Last calculating pen album is combined in pen section variance on four sub spaces, and as stroke distributing equilibrium degree, the more little then pen of visible variance section distributes balanced more with this variance.If after newly importing stroke and joining historical stroke set in first buffer area, the stroke distributing equilibrium degree that calculates surpasses certain pre-set threshold than the ratio stroke distributing equilibrium degree added value of adding the historical stroke in first buffer area before the new input stroke, such as original 0.15 times, the historical stroke in the then definite new input stroke and first buffer area is under the jurisdiction of same character.

B. consider the superposition situation between the historical stroke in the new input stroke and first buffer area, in general, overlapping between the inner stroke of character have a certain rule, less such as the number of new input stroke and historical stroke joining.Therefore, if the degree of overlapping of new input stroke and historical stroke is very high and do not have rule comparatively in a jumble, then it is a new character stroke to a great extent, that is to say, the historical stroke of newly importing in the stroke and first buffer area is under the jurisdiction of kinds of characters.

Such as, calculate the corner dimension that the historical stroke in the new input stroke and first buffer area intersects in twos, if average angle less than predefined thresholding, such as 20 degree, determines that then new stroke and historical stroke are kinds of characters

Certainly, the embodiment of the invention is not limited in above-mentioned judgment mode, can also adopt other judgment modes, gives an example no longer one by one at this.

Step 103, if newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then empty second buffer area and store the first buffer area stored historical stroke into second buffer area, then, empty first buffer area and will newly import stroke and store first buffer area into; Otherwise, will newly import stroke and store first buffer area into.

For can clearly judging the situation that belongs to kinds of characters, can think that the historical stroke in first buffer area has been all strokes of a complete character, therefore, need at first the historical stroke in first buffer area to be dumped to second buffer area, it is discerned to wait for the corresponding identification module in backstage;

For can clearly judging the situation that belongs to same character,, therefore, can directly will should store in first buffer area by new input stroke owing to belong to same character with historical stroke in first buffer area.

Step 104 is discerned according to the two-dimensional coordinate point range of this history stroke of record the historical stroke in described second buffer area, obtains complete character;

In embodiments of the present invention, historical stroke in second buffer area being carried out the identification of character carries out in real time, that is to say, in case when determining that the new historical stroke of importing in the stroke and first buffer area is under the jurisdiction of kinds of characters, promptly trigger the backstage character recognition module processing discerned in the set of the historical stroke that is judged to be complete character in second buffer area.Certainly, character recognition module also can read the set of the historical stroke in second buffer area in real time, discerns processing.

Step 105 shows the complete character that identification obtains.

When discerning, may obtain a plurality of recognition results for each historical stroke set that is judged to be complete character.When showing, first-selected character in the recognition result that generates can be placed the text editing district, and other candidate characters can be preserved by buffer memory or other modes, such as in candidate regions, arrange to show or under editing mode the mode by menu show, be used for the editor corrigendum of subsequent user to recognition result.

This shows, the embodiment of the invention realizes the method for continuous handwriting recognition input, when realizing input continuously by the stack coverage mode in the prior art, adopt the low problem of continuous recognition technology input efficiency of no cutting, by following the tracks of user's stroke input person's handwriting, each stroke is recorded as a two-dimensional coordinate point range separately, in following the tracks of user's input process, by the breakpoint analysis mode the original character cutting of writing space stack is word string independently, and this word string is discerned synchronously.Because the processing of breakpoint analysis and character recognition is carried out synchronously with user's input, promptly in user's input process, determine that new input stroke is gentle and deposit that this carried out character recognition to historical stroke when historical stroke was under the jurisdiction of kinds of characters in the district, thereby avoided the stand-by period of waiting in turn under the existing stack covering input mode, improved input efficiency.

In embodiments of the present invention, the flow process of tracking user stroke input person's handwriting may further comprise the steps as shown in Figure 2:

Step 201 empties each counter and buffer area, and adjacent stroke dead time counter T=0 is set;

Step 202 empties and writes the district, provides the space to give the follow-up input of user;

Step 203 is waited for input;

Step 204 judges whether style of writing starts to write; If then execution in step 205; Otherwise, execution in step 207;

Step 205 is followed the tracks of user's stroke input person's handwriting, and stroke is recorded as a two-dimensional coordinate point range, leaves touch-screen until style of writing; Then adjacent stroke dead time counter T is reset to 0;

Step 206 is input to the back-end processing system with a series of two-dimensional coordinate point ranges that write down;

Step 207, counter T adds 1 with the adjacent stroke dead time;

Step 208, whether the value of adjacent stroke dead time counter T is greater than default dead time threshold value Tr; If then execution in step 209; Otherwise, return step 203;

Step 209 determines to have imported a complete history character, starts the back-end processing system to historical stroke set the carrying out character recognition in second buffer area.

Described back-end processing system is one and integrated the user is imported person's handwriting carry out the system that real-time analysis is handled, and its major function is that the historical person's handwriting to user's input carries out processing such as breakpoint analysis, character recognition, promptly among Fig. 1 step 102 to the process of step 104.That is to say, when the embodiment of the invention realizes the practical application of the method that continuous handwriting recognition is imported, can be applied to two aspects in front-end and back-end of a system respectively.Import person's handwriting in real-time monitoring of front end and collection user stroke, and each stroke is recorded as a two-dimensional coordinate point range separately, the user's stroke that collects is imported processing such as person's handwriting is put in order, identification, import and recognition efficiency thereby can improve in the rear end.

In addition, need to prove, in embodiments of the present invention, when carrying out character recognition, owing to be that the historical stroke set that is judged to be complete character is discerned, rather than prior art is like that direct as previously mentioned extracts feature on original space stack person's handwriting, and recognizer that therefore can integrated various classics has improved discrimination effectively.

Character recognition is carried out to feature in all directions that has very high discrimination such as utilization, and main process is as follows:

(1) pre-service

At first, carry out normalization, be about to the consolidation of character housing to the systemic presupposition size, such as 512*512; To remove away from the sampled point of the main person's handwriting of character then, and carry out smoothing processing, generally be with the alternative current sampling point position coordinates of the average of N point before and after the sampled point; Reduce noise in the person's handwriting signal that collects by modes such as resamplings then.Described resampling is meant that the sampled point sequence resampling with former time equalization is the sampled point sequence of space equalization.

(2) feature extraction

Such as, definition from all directions to be eight directions of the right-falling stroke of left-falling stroke anyhow corresponding to the Chinese-character stroke characteristics.System at first is divided into the character housing N*N word frame (N gets 8 usually).Subsequently the projection vector of projection that makes progress from all directions by the pen section of calculating between neighbouring sample point and accumulation calculating N*N sub-frame person's handwriting and, obtain the proper vector that N*N*8 ties up.Can only it be projected on the nearest both direction the time in pen section between calculating neighbouring sample point, and the projection vector on other directions is counted 0 from all directions to projection.

(3) eigentransformation

The primitive character vector that the N*N*8 that obtains in the step (2) ties up is done to distinguish the property eigentransformation, to reduce dimension and to improve discrimination.Specifically can adopt area of pattern recognition LDA (LinearDiscriminant Analysis commonly used, linear discriminant analysis) eigentransformation method utilizes the LDA transformation matrix of system intialization that the eigenvector of original N*N*8 is reduced to eigenvector (N*N*8) M that M ties up).

(4) carry out pattern match according to language model and character model

Calculate the similarity of each character model in given M dimensional feature vector and the character library, and the character that will have a maximum similarity is as first-selected recognition result.

(5) generate the candidate characters sequence

According to model similarity score ordering in the step (4), several (such as 4) character that score is the highest shows the user as the candidate characters collection.

By above-mentioned character recognition processing procedure, the written handwriting of the original input of user can be mapped as one group of possible ASCII candidate characters, that is to say, when discerning, may obtain a plurality of recognition results for each historical stroke set that is judged to be complete character.When showing, first-selected character in the recognition result that generates can be placed the text editing district, and other candidate characters can be preserved by buffer memory or other modes, such as in candidate regions, arrange to show or under editing mode the mode by menu show, be used for the editor corrigendum of subsequent user to recognition result.Since the embodiment of the invention realize the input of continuous handwriting recognition the method support be continuous handwriting recognition, the semantic information that obtains in the process that the user constantly imports can be enriched constantly, therefore, when carrying out character recognition, can also be in conjunction with language model information, according to follow-up recognition result historical recognition result is carried out in real time corrigendum automatically, further improve discrimination.

Cover in the input mode in the prior art stack,,, mix mutually between the stroke of character, be difficult to distinguish the stroke of historical person's handwriting and current input along with increasing of character because the user imports a plurality of characters writing in the space of setting with the stack coverage mode.The writing of input is difficult to clear representing, thereby the content of user's input clearly can not be represented to the user.

At this problem, the mode that the embodiment of the invention has further taked the person's handwriting desalination to handle makes the user can clearly screen input character, improves user's writing experience, below this is elaborated.

In embodiments of the present invention, further the historical stroke in first buffer area is desalinated processing, and write the real-time historical stroke that shows after new input stroke in first buffer area and desalination are handled in district at screen.Particularly, the mode that historical stroke is desalinated processing has multiple, such as:

(1) according to the rise time sequencing of each stroke in first buffer area each stroke is carried out at the uniform velocity progressive desalination, i.e. the rise time first more desalination of stroke of a specified duration more.

(2) result according to breakpoint analysis desalinates processing to the stroke in first buffer area, specifically comprises:

If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then with described historical stroke to background near disappearing, in this case, the desalination value of historical stroke can be made as 0;

If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of same character, then described historical stroke is normally shown, in this case, the desalination value of historical stroke can be made as 1;

If newly importing the membership of the stroke and the first buffer area stored historical stroke is fringe, then described historical stroke is desalinated demonstration in the watermark mode, in this case, the desalination value of historical stroke can be made as 0.5.

Which kind of need to prove, no matter under above-mentioned desalination processing mode, all can normally show with newly importing stroke.

Above-mentioned (2) kind mode can be adaptive to the speed of user input, and that has avoided that the constant desalination time causes may wait for problem.

As seen, the embodiment of the invention realizes the method for continuous handwriting recognition input, not only can improve input and recognition efficiency effectively, and, handle by desalination, can on small screen device, adopt the mode of stack input, also can make the user clearly screen the character of input historical stroke, when realizing input continuously, improved user's writing experience.

As seen through the above description of the embodiments, those skilled in the art's all or part of step that can be well understood in the foregoing description method can realize by the mode that software adds essential general hardware platform.Based on such understanding, the part that technical scheme of the present invention contributes to prior art in essence in other words can embody with the form of software product, this computer software product can be stored in the storage medium, as ROM/RAM, magnetic disc, CD etc., comprise that some instructions are with so that a computer equipment (can be a personal computer, server, the perhaps network equipment etc.) carry out the described method of some part of each embodiment of the present invention or embodiment.

Correspondingly, the embodiment of the invention also provides a kind of system that realizes continuous handwriting recognition input, as shown in Figure 3, is a kind of structural representation that the embodiment of the invention realizes the system of continuous handwriting recognition input.

In this embodiment, described system comprises:

Data acquisition module 301 is used to follow the tracks of user's stroke input person's handwriting, and each stroke is recorded as a two-dimensional coordinate point range separately;

Breakpoint analysis module 302 is used for determining the new membership of importing the stroke and the first buffer area stored historical stroke;

Stroke sorting module 303, be used for when described breakpoint analysis module 302 determines that the new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters, empty second buffer area and store the first buffer area stored historical stroke into second buffer area, then, also will newly import stroke stores first buffer area into to empty first buffer area; And when described breakpoint analysis module determines that the new input stroke and the first buffer area stored historical stroke are under the jurisdiction of same character, will newly import stroke and store first buffer area into;

Character recognition module 304 is used for the historical stroke of described second buffer area is discerned according to the two-dimensional coordinate point range of this history stroke of record, obtains complete character;

Character display module 305 is used to the complete character that shows that described character recognition module identification obtains.

Need to prove, when character recognition module 304 is discerned for each historical stroke set that is judged to be complete character, may obtain a plurality of recognition results.When 305 pairs of recognition results of character display module show, first-selected character wherein can be placed the text editing district, and other candidate characters can be preserved by buffer memory or other modes, such as in candidate regions, arrange to show or under editing mode the mode by menu show, be used for the editor corrigendum of subsequent user to recognition result.

In embodiments of the present invention, described breakpoint analysis module 302, specifically be used for determining the membership of the new input stroke and the first buffer area stored historical stroke according to the position relation of the new input stroke and the first buffer area stored historical stroke, specifically can comprise: position determination unit, computing unit, first judging unit and second judging unit, wherein:

Described position determination unit is used for determining the start position of new input stroke, and the final position of up-to-date stroke in the set of the first buffer area stored historical stroke;

Described computing unit is used to calculate the distance in described start position and described final position;

Described first judging unit, be used for result of calculation according to described computing unit, during greater than first distance threshold, determine that the described new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters in the distance in described start position and described final position; During less than the second distance threshold value, determine that the described new input stroke and the first buffer area stored historical stroke are to be under the jurisdiction of same character in the distance in described start position and described final position; During more than or equal to the second distance threshold value and smaller or equal to first distance threshold, determine that the membership of the described new input stroke and the first buffer area stored historical stroke is a fringe in the distance in described start position and described final position;

Described second judging unit, be used for when the membership of the described new input stroke and the first buffer area stored historical stroke is fringe, according to the historical stroke set back that new input stroke joins in first buffer area membership of described new input stroke and the first buffer area stored historical stroke is determined in the influence of the equilibrium degree of character stroke on space distribution, perhaps determined the membership of the described new input stroke and the first buffer area stored historical stroke according to the superposition situation between the historical stroke of newly importing in the stroke and first buffer area.

This shows, the embodiment of the invention realizes the system of continuous handwriting recognition input, by following the tracks of user's stroke input person's handwriting, each stroke is recorded as a two-dimensional coordinate point range separately, in following the tracks of user's input process, by the breakpoint analysis mode the original character cutting of writing space stack is word string independently, and this word string is discerned synchronously.Because the processing of breakpoint analysis and character recognition is carried out synchronously with user's input, promptly in user's input process, determine that new input stroke is gentle and deposit that this carried out character recognition to historical stroke when historical stroke was under the jurisdiction of kinds of characters in the district, thereby avoided the stand-by period of waiting in turn under the existing stack covering input mode, improved input efficiency.

As shown in Figure 4, be the another kind of structural representation that the embodiment of the invention realizes the system of continuous handwriting recognition input.

With embodiment illustrated in fig. 3 different be that in this embodiment, described system further comprises:

Desalination processing module 401 is used for the historical stroke of first buffer area is desalinated processing;

Write display module 402, be used for writing district new input stroke that shows first buffer area in real time and the historical stroke of desalinating after handling.

When concrete the application, described desalination processing module 401 can have multiple implementation.

Can carry out at the uniform velocity progressive desalination to described historical stroke according to the rise time sequencing of each stroke in first buffer area such as, desalination processing module 401.

For another example, desalination processing module 401 can also be desalinated processing to the historical stroke in first buffer area in such a way:

If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then with described historical stroke to background near disappearing;

If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of same character, then described historical stroke is normally shown;

If newly importing the membership of the stroke and the first buffer area stored historical stroke is fringe, then described historical stroke is desalinated demonstration in the watermark mode.

The system of the embodiment of the invention can be applied to small-sized touch panel device, satisfy the demand of the continuous handwriting recognition input of user, in the limited input that superposes in the district of writing, handle by desalination historical stroke, can make the user clearly distinguish the stroke of historical person's handwriting and current input, the content of user's input is clearly represented to the user, improve user's writing experience.

The embodiment of the invention realizes the method and system of continuous handwriting recognition input, goes for following two kinds of situations at least:

1) smart machine is equipped with touch pad (being generally the peripheral facility of equipment, for example handwriting pad etc.), and the fixed area of user on touch pad write repeatedly;

2) smart machine is equipped with touch-screen (being generally the equipment of the system integration), but the user can write repeatedly in certain fixed area of touch-screen.

Below by further describing the embodiment of the invention for example.

Suppose user's input " adult " two Chinese characters continuously on touch panel device, stroke from left to right promptly shown in Figure 5 then utilizes the embodiment of the invention, and the demonstration that the input person's handwriting is write the district at screen as shown in Figure 6.

Below Fig. 5 is elaborated to the process of Fig. 6, mainly comprises:

1. when the user imports the first stroke from left to right, owing to be the first stroke, be under the jurisdiction of same character so the breakpoint analysis module is judged itself and historical stroke, the desalination processing module is set it and is shown for normal.

2. when the user imports second, the breakpoint analysis module is determined the membership of second and the first stroke.Determine that second and the first stroke are under the jurisdiction of same character, therefore desalinating processing module also is set at normal demonstration with second.

3. when the user imported the 3rd, the breakpoint analysis module was determined the 3rd and the membership of the first two, and is judged as the fringe situation, the person's handwriting of at this moment front being imported desalination demonstration, as shown in Figure 6.

Therefore 4. when the user imported the 4th, the breakpoint analysis module determined that the 3rd and the 4th belongs to two characters respectively, three of the first strokes to the was removed from screen or was shown in the watermark mode.Provided the example that shows in the watermark mode among Fig. 6.

5. when the user imported the 5th, the breakpoint analysis module judged that the 4th and the 5th is under the jurisdiction of same character, so person's handwriting normally shows.

In said process,,, think that then the stroke that next will write belongs to different characters respectively with the stroke of having imported as surpassing 0.5 second if lift an overlong time.

Each embodiment in this instructions all adopts the mode of going forward one by one to describe, and identical similar part is mutually referring to getting final product between each embodiment, and each embodiment stresses all is difference with other embodiment.Especially, for system embodiment, because it is substantially similar in appearance to method embodiment, so describe fairly simplely, relevant part gets final product referring to the part explanation of method embodiment.System embodiment described above only is schematic, wherein said module and unit as the separating component explanation can or can not be physically to separate also, such as, data acquisition module in the system can be deployed in the front end of equipment, and other modules in the system are deployed in the rear end of equipment, make its synchronous operation, realize continuous handwriting recognition input.In addition, can also select wherein some or all of module to realize the purpose of present embodiment scheme according to the actual needs.Those of ordinary skills promptly can understand and implement under the situation of not paying creative work.

More than disclosed only be preferred implementation of the present invention; but the present invention is not limited thereto; any those skilled in the art can think do not have a creationary variation, and, all should drop in protection scope of the present invention not breaking away from some improvements and modifications of being done under the principle of the invention prerequisite.

Claims (10)

1. a method that realizes continuous handwriting recognition input is characterized in that, comprising:
Follow the tracks of user's stroke input person's handwriting, and each stroke is recorded as a two-dimensional coordinate point range separately;
Determine the membership of the new input stroke and the first buffer area stored historical stroke;
If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then empty second buffer area and store the first buffer area stored historical stroke into second buffer area, then, also will newly import stroke stores first buffer area into to empty first buffer area;
Otherwise, will newly import stroke and store first buffer area into;
Historical stroke in described second buffer area is discerned according to the two-dimensional coordinate point range of this history stroke of record, obtained complete character;
Show the complete character that identification obtains.
2. the method for claim 1 is characterized in that, the membership of the described definite new input stroke and the first buffer area stored historical stroke comprises:
Position relation according to stored historical stroke in the new input stroke and first buffer area is determined the new membership of importing the stroke and the first buffer area stored historical stroke, specifically comprises:
Determine the start position of new input stroke, and the final position of up-to-date stroke in the set of the first buffer area stored historical stroke;
If the distance in described start position and described final position, determines then that the described new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters greater than first distance threshold;
If the distance in described start position and described final position, determines then that the described new input stroke and the first buffer area stored historical stroke are to be under the jurisdiction of same character less than the second distance threshold value;
If the distance in described start position and described final position, determines then that the membership of the described new input stroke and the first buffer area stored historical stroke is a fringe more than or equal to the second distance threshold value and smaller or equal to first distance threshold;
If the membership of the described new input stroke and the first buffer area stored historical stroke is a fringe, then the membership of described new input stroke and the first buffer area stored historical stroke is determined to the influence of the equilibrium degree of character stroke on space distribution in the historical stroke set back that joins in first buffer area according to new input stroke, perhaps determines the membership of the described new input stroke and the first buffer area stored historical stroke according to the superposition situation between the historical stroke of newly importing in the stroke and first buffer area.
3. method as claimed in claim 2 is characterized in that, described method also comprises:
Historical stroke in first buffer area is desalinated processing;
Writing the real-time historical stroke that shows after new input stroke in first buffer area and desalination are handled in district.
4. method as claimed in claim 3 is characterized in that, described historical stroke in first buffer area is desalinated to handle comprise:
According to the rise time sequencing of each stroke in first buffer area described historical stroke is carried out at the uniform velocity progressive desalination.
5. method as claimed in claim 3 is characterized in that, described historical stroke in first buffer area is desalinated to handle comprise:
If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then with described historical stroke to background near disappearing;
If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of same character, then described historical stroke is normally shown;
If newly importing the membership of the stroke and the first buffer area stored historical stroke is fringe, then described historical stroke is desalinated demonstration in the watermark mode.
6. a system that realizes continuous handwriting recognition input is characterized in that, comprising:
Data acquisition module is used to follow the tracks of user's stroke input person's handwriting, and each stroke is recorded as a two-dimensional coordinate point range separately;
The breakpoint analysis module is used for determining the new membership of importing the stroke and the first buffer area stored historical stroke;
The stroke sorting module, be used for when described breakpoint analysis module determines that the new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters, empty second buffer area and store the first buffer area stored historical stroke into second buffer area, then, also will newly import stroke stores first buffer area into to empty first buffer area; And when described breakpoint analysis module determines that the new input stroke and the first buffer area stored historical stroke are under the jurisdiction of same character, will newly import stroke and store first buffer area into;
Character recognition module is used for the historical stroke of described second buffer area is discerned according to the two-dimensional coordinate point range of this history stroke of record, obtains complete character;
The character display module is used to the complete character that shows that described character recognition module identification obtains.
7. system as claimed in claim 6, it is characterized in that, described breakpoint analysis module specifically is used for the membership according to the definite new input stroke of the position relation of the new input stroke and the first buffer area stored historical stroke and the first buffer area stored historical stroke, comprising:
Position determination unit is used for determining the start position of new input stroke, and the final position of up-to-date stroke in the set of the first buffer area stored historical stroke;
Computing unit is used to calculate the distance in described start position and described final position;
First judging unit is used for the result of calculation according to described computing unit, during greater than first distance threshold, determines that the described new input stroke and the first buffer area stored historical stroke are under the jurisdiction of kinds of characters in the distance in described start position and described final position; During less than the second distance threshold value, determine that the described new input stroke and the first buffer area stored historical stroke are to be under the jurisdiction of same character in the distance in described start position and described final position; During more than or equal to the second distance threshold value and smaller or equal to first distance threshold, determine that the membership of the described new input stroke and the first buffer area stored historical stroke is a fringe in the distance in described start position and described final position;
Second judging unit, be used for when the membership of the described new input stroke and the first buffer area stored historical stroke is fringe, according to the historical stroke set back that new input stroke joins in first buffer area membership of described new input stroke and the first buffer area stored historical stroke is determined in the influence of the equilibrium degree of character stroke on space distribution, perhaps determined the membership of the described new input stroke and the first buffer area stored historical stroke according to the superposition situation between the historical stroke of newly importing in the stroke and first buffer area.
8. system as claimed in claim 7 is characterized in that, described system also comprises:
The desalination processing module is used for the historical stroke of first buffer area is desalinated processing;
Write display module, be used for writing district new input stroke that shows first buffer area in real time and the historical stroke of desalinating after handling.
9. system as claimed in claim 8 is characterized in that,
Described desalination processing module specifically is used for according to the rise time sequencing of first each stroke of buffer area described historical stroke being carried out at the uniform velocity progressive desalination.
10. system as claimed in claim 8 is characterized in that,
Described desalination processing module specifically is used in such a way the historical stroke of first buffer area being desalinated processing:
If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of kinds of characters, then with described historical stroke to background near disappearing;
If newly import stroke and the first buffer area stored historical stroke is under the jurisdiction of same character, then described historical stroke is normally shown;
If newly importing the membership of the stroke and the first buffer area stored historical stroke is fringe, then described historical stroke is desalinated demonstration in the watermark mode.
CN 201110076004 2011-03-28 2011-03-28 Method and system for realizing continuous handwriting recognition input CN102156577B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201110076004 CN102156577B (en) 2011-03-28 2011-03-28 Method and system for realizing continuous handwriting recognition input

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201110076004 CN102156577B (en) 2011-03-28 2011-03-28 Method and system for realizing continuous handwriting recognition input

Publications (2)

Publication Number Publication Date
CN102156577A true CN102156577A (en) 2011-08-17
CN102156577B CN102156577B (en) 2013-05-29

Family

ID=44438094

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201110076004 CN102156577B (en) 2011-03-28 2011-03-28 Method and system for realizing continuous handwriting recognition input

Country Status (1)

Country Link
CN (1) CN102156577B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102508598A (en) * 2011-10-09 2012-06-20 北京捷通华声语音技术有限公司 Method and device for gradually blanking character strokes
CN102521577A (en) * 2011-12-20 2012-06-27 安徽科大讯飞信息科技股份有限公司 Handwriting recognition, synthesis and tracking method of interactive multimedia device
CN103390013A (en) * 2012-05-11 2013-11-13 株式会社东芝 Electronic device and handwritten document processing method
WO2015161823A1 (en) * 2014-04-25 2015-10-29 夏普株式会社 Handwriting recognition method and device
CN107219935A (en) * 2017-05-25 2017-09-29 哈尔滨工业大学 It is a kind of towards continuous writing Chinese character, support interaction Chinese character input system and method

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101561725A (en) * 2008-04-16 2009-10-21 汉王科技股份有限公司 Method and system of fast handwriting input
CN101964048A (en) * 2010-07-19 2011-02-02 安徽科大讯飞信息科技股份有限公司 Character recognition method and system

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101561725A (en) * 2008-04-16 2009-10-21 汉王科技股份有限公司 Method and system of fast handwriting input
CN101964048A (en) * 2010-07-19 2011-02-02 安徽科大讯飞信息科技股份有限公司 Character recognition method and system

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102508598A (en) * 2011-10-09 2012-06-20 北京捷通华声语音技术有限公司 Method and device for gradually blanking character strokes
CN102508598B (en) * 2011-10-09 2014-03-05 北京捷通华声语音技术有限公司 Method and device for gradually blanking character strokes
CN102521577A (en) * 2011-12-20 2012-06-27 安徽科大讯飞信息科技股份有限公司 Handwriting recognition, synthesis and tracking method of interactive multimedia device
CN103390013A (en) * 2012-05-11 2013-11-13 株式会社东芝 Electronic device and handwritten document processing method
WO2015161823A1 (en) * 2014-04-25 2015-10-29 夏普株式会社 Handwriting recognition method and device
CN105095924A (en) * 2014-04-25 2015-11-25 夏普株式会社 Handwriting recognition method and device
CN107219935A (en) * 2017-05-25 2017-09-29 哈尔滨工业大学 It is a kind of towards continuous writing Chinese character, support interaction Chinese character input system and method

Also Published As

Publication number Publication date
CN102156577B (en) 2013-05-29

Similar Documents

Publication Publication Date Title
US20160283579A1 (en) Information processing device, information processing method, and computer program
US10032483B2 (en) Summary image browsing system and method
CN104662491B (en) Automatic gesture for sensing system is recognized
US9880655B2 (en) Method of disambiguating water from a finger touch on a touch sensor panel
CN102645972B (en) The movement of computing equipment is used to strengthen the explanation of the incoming event produced when mutual with this computing equipment
US8368723B1 (en) User input combination of touch and user position
RU2650029C2 (en) Method and apparatus for controlling application by handwriting image recognition
US8537129B2 (en) Techniques for recognizing movement of one or more touches across a location on a keyboard grid on a touch panel interface
CN104246679B (en) Gesture input at dummy keyboard
CN105183211B (en) For managing the method and relevant device of the software application opened parallel
CN101739207B (en) Mobile terminal having touch screen and method for displaying tag information therof
CN101324828B (en) Method and device for operating multi-point touch enabled device
KR101675178B1 (en) Touch-screen user interface
CN1758205B (en) Flick gesture
CN105247540B (en) Manage real-time handwriting recognition
US6249606B1 (en) Method and system for gesture category recognition and training using a feature vector
CN101563666B (en) User interface device
KR20170046131A (en) Detecting selection of digital ink
CN102460362B (en) Glyph entry on computing device
CN103412882B (en) A kind of method and device identifying that consumption is intended to
CN101611417B (en) Method for character recognition
CN104252264B (en) Information processor and control method
CN102349040B (en) For comprising the system and method at the interface of the haptic effect based on surface
CN102622127B (en) Touch contact based on prediction is followed the tracks of
JP3267047B2 (en) Information processing device by voice

Legal Events

Date Code Title Description
PB01 Publication
C06 Publication
SE01 Entry into force of request for substantive examination
C10 Entry into substantive examination
GR01 Patent grant
C14 Grant of patent or utility model
CP03 Change of name, title or address

Address after: Wangjiang Road high tech Development Zone Hefei city Anhui province 230088 No. 666

Patentee after: Iflytek Co., Ltd.

Address before: 230088 No. 616, Mount Huangshan Road, hi tech Development Zone, Anhui, Hefei

Patentee before: Anhui USTC iFLYTEK Co., Ltd.

C56 Change in the name or address of the patentee

Owner name: IFLYTEK CO., LTD.

Free format text: FORMER NAME: ANHUI USTC IFLYTEK CO., LTD.